A secure, extensible platform for AI services including document processing, agents, and more. Built with FastAPI to provide a scalable architecture for AI-powered features.

• Document Processing: Extract structured content from PDFs, Word documents, and more
• AI Agents: (Coming Soon) Autonomous agents for specialized tasks
• Processing Chains: (Coming Soon) Multi-step AI workflows
• Knowledge Indexing: (Coming Soon) Lemma and concept indexing

# Clone the repository
git clone <repository-url>
cd api-proxy

# Install dependencies
pip install -r requirements.txt

# Set up environment variables
cp .env.example .env
# Edit .env to add your API_KEY

# Run the development server
python main.py

Create a .env file with these variables:

# Required for API authentication
API_KEY=your_api_key_here

# Optional (defaults shown)
DOCLING_API_URL=http://docling-serve-cpu.railway.internal:3000
DOCLING_SERVICE_NAME=docling-serve-cpu
DOCLING_SERVICE_PORT=3000

Base URL: https://ai.g2i.co/api/v1

• Health Check: GET /health - Verify the service is running
• API Documentation: GET /docs - Interactive API documentation
• Agents (Coming Soon): GET /agents - List available AI agents

All document processing endpoints require authentication with:

Authorization: Bearer <api_token>

• Convert URL: POST /document/convert/source - Process documents from URLs
• Convert File: POST /document/convert/file - Process uploaded document files
• Async Processing: POST /document/convert/source/async - Start async document conversion
• Check Status: GET /document/status/poll/{task_id}?wait={seconds} - Poll for status updates
• Get Results: GET /document/result/{task_id} - Retrieve conversion results

# Process a document from URL
curl -X 'POST' \
  'https://ai.g2i.co/api/v1/document/convert/source' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer <api_token>' \
  -d '{
    "http_sources": [{"url": "https://arxiv.org/pdf/2206.01062.pdf"}],
    "options": {
      "to_formats": ["json"],
      "from_formats": ["pdf"],
      "image_export_mode": "embedded",
      "do_ocr": true
    }
  }' \
  --output 'output.zip'

# List available agents
curl -X 'GET' 'https://ai.g2i.co/api/v1/agents'

For a complete list of options, see the API documentation.

app/
├── api/              # API routes by version
├── core/             # Configuration and utilities
├── middleware/       # Authentication middleware
├── models/           # Data models/schemas
├── services/         # Service integrations
├── utils/            # Utility functions
└── app.py            # Application entry point

The application uses Railway Nixpacks for deployment. Configuration is in railway.json.